CRF-like immunoreactivity selectively labels preganglionic sudomotor neurons in cat

253

Brain Research, 599 (1992) 253-260 © 1992 Elsevier Science Publishers B.V. All rights reserved 0006-8993/92/$05.00

BRES 18378

CRF-like immunoreactivity selectively labels preganglionic sudomotor neurons in cat A.D. Shafton, B.J. O l d f i e l d a n d R . M . M c A l l e n The Howard FIorey Institute of Experimental Physiology and Medicine, Unicersity of Melbourne, Parkville, Vic. (Australia) (Accepted 28 July 1992)

Key words: Corticotrophin releasing factor; Peptide; Sympathetic neuron; Preganglionic; Chemical coding; Sweat gland

Immunohistochemical and neuronal tracing methods were used in cats to determine which type of postganglionic sympathetic neuron is innervated by preganglionic neurons which contain corticotrophin releasing factor-like immunoreactivity (CRF-LI). Preganglionic neurons with CRF-LI have their cell bodies at two restricted levels of the spinal cord and terminate in the stellate and lower lumbar ganglia. CRF-LI terminal baskets in stellate and lumbar ganglia surrounded cell bodies, 96-99% of which showed no tyrosine hydroxylase (TH)-LI (presumptive cholinergic neurons). Calcitonin gene-related peptide (CGRP)-LI was used to label the cholinergic ganglion cells which innervate sweat glands: 96-99% of those were confirmed as lacking TH-LI, while the remainder showed weak staining. Every one of over 6000 CRF-LI terminal baskets counted in 4 stellate and 6 lumbar ganglia was found to surround a cell body with CGRP-LI; conversely, 81-86% of the cell bodies showing CGRP-LI were surrounded by CRF-LI terminal baskets. In 3 cats, the retrograde tracer fluorogold was used to label postganglionic neurons projecting to the paw pads (a population which includes both cholinergic sudomotor neurons and noradrenergic vasoconstrictor neurons). Between 26 and 38% of the retrogradely labelled ganlion cells were surrounded by CRF-LI terminal baskets. We conclude that in cats, preganglionic sympathetic neurons with CRF-LI are sudomotor in function.

INTRODUCTION A variety of peptides coexist with classical neurotransmitters in different sympathetic pre- and postganglionic neurons 2-7'9't°'12-17'2°-25. Frequently, the neurons of each functional type contain a characteristic peptide or combination of peptides. Thus, immunohistochemical staining patterns can be used to identify the neurons of a particular functional class (although the pattern shown by homologous neurons may differ between species) 7. This principle of chemical coding for function has been well demonstrated to hold among enteric and post ganglionic neurons 7, but few equivalent data exist for preganglionic sympathetic neurons. Such data would be valuable as a basis for tracing central autonomic control pathways. K r u k o f f 12 showed that in cats a subgroup of preganglionic neurons contains corticotrophin releasing factor-like immunoreactivity (CRF-LI), and that these are found in two distinct spinal levels: the first between thoracic segments 2 and 7 with a peak density at T4,

the second in lumbar segments 2 and 3. A possible explanation for this selective distribution could be that CRF-LI labels a class of preganglionic neuron whose functional target is restricted to the limbs. One likely candidate system is therefore the sudomotor supply, which in cats only innervates sweat glands in the paw pads 11'18'19. The postganglionic neurons of this pathway are unusual, in that they are cholinergic 11'2°. In cat sympathetic ganglia, there exist two populations of cholinergic neurons: those which supply the sweat glands and those which form the cholinergic vasodilator supply to skeletal muscle 2°. While both groups contain acetylcholine and vasoactive intestinal peptide-like immunoreactivity (VIP-LI), the sudomotor neurons can be distinguished by their additional content of calcitonin gene-related peptide- and substance P-like immunoreactivities (CGRP-LI, SP-LI) 2°. The present experiments were undertaken to test whether CRF-LI-containing preganglionic neurons selectively innervate cholinergic postganglionic neurons. We then sought to determine whether those postgan-

Correspondence: R.M. McAllen, Howard Florey Institute, University of Melbourne, Parkville, Vic. 3052, Australia. Fax: (61) (3)-348-1702.

254 gtionic neurons belonged to the sudomotor supply, firstly by staining them for CGRP-LI, and secondly by retrogradely labelling them with a neuronal tracer injected into the paw pads. Some of this work has been published in abstract form 25.

MATERIALS AND METHODS

Stellate and lumbar sympathetic ganglia were dissected post mortem from adult cats of either sex. Seven of the cats were killed with an overdose of pentobarbital sodium (120 mg/kg i.v.) following an acute physiological experiment which did not involve the sympathetic ganglia. Sixteen other cats were directly anaesthetised with pentobarbital sodium (60 mg/kg i.p.). All but 4 of these cats were perfused transcardially with 1 1 saline followed by 2 1 4% paraformaldehyde in 0.1 M sodium phosphate buffer (pH 7.2). Ganglia dissected from the remaining 4 cats were fixed by immersion overnight in Bouin's solution; 3 of these were previously desheathed by dissections . Ganglia were then transferred to 20% sucrose in phosphate buffer for at least 4 h. The different modes of preparation gave indistinguishable results, and data from them are hereafter considered together. Retrograde tracing experiments were performed on 3 further cats, and these were anaesthetised intraperitoneally with ketamine hydrochloride (44 mg/kg) and diazepam (2.5 mg). One or 2 forepaws were injected with 5-7% fluorogold (Fluorochrome Inc., Englewood, CO, U.S.A), 50-350 #.1 per paw, distributed amongst the pads. They recovered uneventfully and were left for 10-14 days before being re-anaesthetised and perfusion-fixed as described above.

Antibodies and histochemi~ al procedure.s Sections of 15 30 /zm were cut longitudinally from ganglia using a freezing microtome, washed in phosphate buffer and placed in 10% normal horse serum for 1 h. then immersed in primary antibody solution for 48 h. After washing 3 times in phosphate buffer, sections were exposed to secondary antibody solution for 45 min. The primary antibody solutions were diluted in phosphate buffer containing 0.3% Triton X-100 and 2% normal horse serum. The primary antibodies used were: (1) mouse monockmal anti-TH (lncstar Corp., Stillwater, MN, USA; dilution 1:2000 or 1:5(100): (2) rabbit polyclonal anti-sheep CRF (kindly donated by Dr. W. Vale, Salk Institute, San Diego, CA, USA: dilution 1:400 or I: ](X10); (3) rabbit polyclonal anti-CGRP (Amersham Australia Pty.. Melbourne, Vic., Australia; dilution 1 : 1000). Secondary antibody solutions were diluted 1:2(10 in phosphate buffer containing 2% normal horse serum. Those used were: (1) rhodamine-tagged goat anti-rabbit IgG (Caltag Labs. San Francisco, CA, USA); (2) Texas Red-tagged donkey anti-rabbit lgG (Jackson Labs., West Grove, PA, USA); (3) biotinylated horse anti-mouse IgG (Vector Labs., Burlingame, CA, USA); (4) biotinylated goat anti-rabbit IgG (Vector). The biotinylated secondary antibodies (3 and 4) were localised by reaction for 45 rain with avidin-conjugated fluoresceine isothiocyanate (FITC. Vector; dilution 1:100).

Specificity of primary antibodies The primary antibodies used here have been welt characterised for specificity. The monoclonal mouse antibody to TH recognises an epitope in the mid-portion of the TH molecule and does not crossreact with dopamine /3 hydroxylase, phenylethanolamine N-methyl transferase, phenylalanine hydroxylase or tryptophan hydroxylase (Incstar). The staining associated with the rabbit polyclonal antibody to CRF could be completely abolished by preincubation with CRF

Fig. 1. Paired views of a section of stellate ganglion double-stained for CRF-LI (Texas red) (a) and TH-LI (FITC) (b). Note that CRF-LI terminal baskets (arrowheads) surround ganglion cells which lack TH-LI (arrowed). Bar = 25/xm.

255

Fig. 2. Paired views of a section of stellate ganglion double-stained for CGRP-LI (Texas red) (a) and TH-LI (FITC) (b). Note that the arrowed cell bodies show CGRP-LI but lack TH-LI. Bar = 25/~m.

peptide 26. The polyclonal rabbit antibody to CGRP does not crossreact with other peptides (Amersham). However, some antibodies to CGRP have been found to crossreact with an unidentified cellular component 9. Accordingly, rat CGRP (Peninsula Labs., Belmont, CA, USA) was mixed to a concentration of 1 /.tM into a working dilution of CGRP antibody for 3 h before it was applied in the normal way to sections of stellate ganglion. The effects of this preabsorption on the staining patterns seen are described in the results section.

Double staining procedures For double staining when the primary antibodies were raised in different species (staining for TH-LI with either CRF-LI or CGRPLI), both primary antibodies and later both secondary antibodies were applied together. When CRF-LI and CGRP-LI staining were combined (both primary antibodies raised in rabbit), the whole

immunostaining procedure was first completed for CRF-LI and then repeated for CGRP-LI. TH-LI was always visualised using the biotinylated horse anti-mouse secondary antibody (3) and avidin-FITC. When either CRF-LI or CGRP-LI staining were to be combined with TH-LI, they were visualised by the use of rhodamine- or Texas red-conjugated secondary antibodies (1,2). When CRF-LI staining was combined with either CGRP-LI or fluorogold fluorescence, it was visualised using biotinylated goat anti-rabbit secondary antibody (4) and avidin-FITC, the CGRP-LI being visualised with Texas red as described above. Rhodamine, used as a fluorophore in early experiments, was replaced by the superior Texas red in later experiments. After final washes, sections were mounted on coated glass slides, and coverslipped with glycergel (DAKO Corp., Carpinteria, CA, USA). Sections were examined on a Leitz Dialux microscope, alter-

Fig. 3. Paired views of a section of stellate ganglion double-stained for CRF-LI (FITC) (a) and CGRP-LI (Texas red) (b), showing a CRF-LI terminal basket (arrowheads) surrounding a cell body with CGRP-LI (arrowed). Bar = 25/xm.

256 nately using the filter blocks for rhodamine, FITC and where appropriate, fluorogold, using a 25 × or 50 × water immersion objective.

TABLE 1

Ganglion Experiments Stellate ganglia from 9 cats and lumbar ganglia 5, 6 and 7 from 3 cats were double stained for CRF-LI and TH-LI. Stellate ganglia from 4 cats and lumbar ganglia from 3 cats were double-stained for TH-LI and CGRP-LI. Stellate ganglia from 4 cats and lumbar ganglia from 3 cats were double-stained for CRF-LI and CGRP-LI. Stellate ganglia from 3 cats which had been retrogradely labelled with fluorogold, were stained for CRF-LI.

RESULTS

TH-L1 staining The large majority of cell bodies in the stellate and lumbar ganglia showed TH-LI 2°, though some were stained more weakly than others. Each cell showed an even fluorescence throughout its cytoplasm. A separate subgroup of ganglion cells was seen, in which TH-LI was absent (Figs. lb and 2b). Fibres of passage with TH-LI were also seen, though none of these appeared to form terminal specialisations around ganglion cells. CRF-LI staining No cell body showing CRF-LI was found in either the stellate or lumbar ganglia, but fibres with CRF-LI were widespread in both. Fibres showing CRF-LI fell into two morphological types: continuous, wavy fibres of passage coursing between cells, and arrays of medium-bright boutons surrounding ganglion cells ('CRF-LI terminal baskets'; Figs. la and 3a). In earlier experiments we used the CRF antibody at a dilution of 1:1000; this was the case for all experiments where CRF-LI was compared with TH-LI (see below). In all other experiments we used a dilution of 1:400, because we found that this substantially increased the yield of identifiable CRF-LI terminal baskets. To minimise the effect of understaining on the numerical analysis, conjunctions between CRF-LI terminal baskets and any cell type are expressed as a percentage of the total number of CRF-LI terminal baskets found in that ganglion, unless stated otherwise. Combined TH-LI and CRF-LI staining In each of 9 stellate ganglia, 182-240 CRF-LI terminal baskets were found surrounding individual ganglion cells (CRF antibody dilution 1:1000 in all cases). In every ganglion, between 96 and 99% of the cells which were surrounded by CRF-LI terminal baskets were negative for TH-LI (see Table I). The TH-LI staining in the other 1 - 4 % of cells surrounded by CRF-LI terminal baskets was always weak. A total of 233 CRF-LI terminal baskets were counted from 9 lumbar ganglia (L 5, L 6 and L 7 taken

Stellate (n = 9)

L5 (n = 3) L6 (n = 3) L7 (n = 3)

Total CRF-LI terminal baskets *

CRL-LI terminal baskets around cells lacking TH-LI

1946 182-243 per ganglion

19(14 (97.8%] 178 -240 (96-99%) per gangion

3

3 (lt10%)

109

106 (97c~)

121

117 ( 9 7 ~ )

* C R F antibody dilution 1 : I 000 all cases.

from each of 3 cats). Ninety seven percent of the cells surrounded by CRF-LI terminal baskets were negative . for TH-LI, while the remaining 3% showed weak TH-LI (Table I).

CGRP-LI staining CGRP-LI was found in a minority of stellate and lumbar ganglion cells. The principle staining pattern consisted of a bright granular appearance against a smooth background staining throughout the cytoplasm (Figs. 2a and 3b). This pattern disappeared after preabsorption with C G R P (see Materials and Methods) and was accepted as showing genuine CGRP-LI. As have previous workers 9. we sometimes found artefactual staining patterns with the C G R P antibody. These took the form of two distinct patterns: z patchy perinuclear stain and a weak. localised patch at one pole of the cell. Both of the latter staimng patterns survived preabsorption with CGRP. and were therefore considered artefactual and will not be dealt with further 9. Cells with CGRP-LI were found mostly in one region of any section, and were common only in a minority of sections of any one ganglion. They were usually surrounded by unstained neurons~ and never formed local clusters of more than 3 cells. They thus followed the 'scattered' distribution described by Lindh et al. z°. A few fibres showing CGRP-LI were also seen coursing between ganglion cells, but no terminal specialisations were seen. Combined CGRP-LI and TH-LI staining In four stellate ganglia from 4 cats, a total of 8496 neurons with CGRP-LI were counted, Of these, 261 (3%) showed weak TH-LI, while the remaining 97% showed none (Fig. 2). A total of 343 cell bodies with CGRP-LI were counted in 9 lumbar ganglia taken from 3 cats; 7 of these cell bodies (2%) showed weak TH-LI, but the remainder showed nonc (Table Ilk

257 T A B L E II

Ganglion

Stellate (n = 4) L5 (n = 3) L6 (n = 3) L7 (n = 3)

Total cell bodies with CGRP-LI

Cell bodies with CGRP-LI but lacking TH-LI

bodies showing CGRP-LI were surrounded by CRF-LI terminal baskets (see Table III). Similar proportions but lower numbers were found in the L 6 and L 7 ganglia taken from 3 cats (Table III).

8496

8235 (96.9%)

Retrograde labelling with fluorogold

6

6 (100%)

145

143 (99%)

192

188 (98%)

Combined CRF-LI and CGRP-LI staining In 4 stellate ganglia from 4 cats totals of 8061 cell bodies showing CGRP-LI and 6735 CRF-LI terminal baskets were counted (CRF antibody dilution 1:400 in all cases). Without exception, CRF-LI terminal baskets were always found surrounding cell bodies with CGRPLI (Fig. 3). Conversely, between 81 and 86% of the cell

Fluorogold injected into the forepaw pads retrogradely labelled small numbers of stellate ganglion cells in 3 cats. In one cat the injections were bilateral, and neurons in both stellate ganglia were labelled, but none in 3 control lumbar ganglia. In 2 cats where only one forepaw was injected, neurons were labelled only in the ipsilateral, not the contralateral stellate ganglion. In all ganglia, bright labelling was seen in capillary endothelial ceils.

Combined retrograde labelling with CRF-LI staining The fluorescence intensity of fluorogold-labelled neurons was considerably reduced by the immunohisto-

T A B L E III

Ganglion

Stellate (n = 4)

L5 (n = 3) L6 (n = 3) L7 (n = 3)

Total CRF-LI terminal baskets *

CRF-LI terminal baskets around cells with CGRP-L1

Total cells with CGRP-LI

Cells with CGRP-LI surrounded by CRF-LI terminal baskets

6735 1 6 3 4 - 1 7 5 6 per ganglion

6735 (100%) (100%)

8061 1905-2114 per ganglion

6735 (83.6%) 1634-1 756 (81-86%) per ganglion

0

0

0

0

111

111 (100%)

134

111 (83%)

156

156 (100%)

185

156 (84%)

* C R F antibody dilution 1 : 400 in all cases.

Fig. 4. a: shows a section of stellate ganglion including a cell body (arrowed) which was retrogradely labelled by fluorogold previously injected into the paw pads. b: when viewed using filters for FITC fluorescence, the same cell can be seen surrounded by a CRF-LI terminal basket (arrowheads). (Some fluorogold fluorescence breaks through the FITC filters.) Bar = 25 ~tm.

258 'FABLE IV

Ganglion

Retrogradely labelled cells'

Retrogradely labelled cells surrounded by CRF-LI terminal baskets *

Stellate (n = 4)

2374 692-963 per ganglion

796 (34%) 179-366 (26-38%) per ganglion

• CRF antibody dilution 1:400 in all cases.

chemical procedure, when compared with untreated sections. Nevertheless, totals of 719, 963 and 692 retrogradely labelled neurons were identified in 3 stellate ganglia which had also been stained for CRF-LI (antibody dilution 1:400). Of those retrogradely labelled neurons, 251,366, and 179, respectively, were found to be surrounded by CRF-LI terminals (35, 38 and 26%; Fig. 4, Table 4). DISCUSSION The results presented here show that, in cat stellate and lumbar ganglia, CRF-LI-containing preganglionic terminals selectively surround postganglionic sudomotor neurons. The supporting evidence for this conclusion is as follows: (1) approximately 98% of CRF-LI terminal baskets were found to surround ganglion cells which lacked TH-LI (presumptive cholinergic neurons). (2) All ganglion cells surrounded by CRF-LI terminal baskets showed CGRP-LI: such cells have previously been identified as sudomotor neurons z°. (3) About one third of the ganglion cells retrogradely labelled from the paw pad, (a population which includes noradrenergic cutaneous vasoconstrictor neurons as well as sudomotor neurons), were surrounded by CRF-LI terminal baskets. Although CRF-LI terminal baskets were always found surrounding ganglion cells with CGRP-LI (i,e. presumptive sudomotor neurons), only 84% of such cells were found to have CRF-LI terminal baskets around them. The robust CGRP-LI staining made it easy to distinguish between stained and unstained cells; but as indicated above, it is quite possible that not all CRF-LI terminal baskets were effectively staLned, even with the highest antibody concentration used. It may be that full staining would reveal that all ~ n g ~ i c sudomotor neurons are innervated by CRF-LI t e r m ~ a l baskets. Alternatively, perhaps not all pregangtionic sudomotor neurons express CRF-LI, at least to levels detectable by these methods. In this study, approximately 2% of CRF-LI terminal baskets were found to innervate cells which showed

some TH-LI, although in every, case the I-H-LI staining was quite weak. Similarly, around 3q~ of cells showing CGRP-LI also showed some (weak) TH-LI. These few exceptions to the general rule can be explained if the cells in question are indeed cholinergic sudomotor neurons which have retained somc noradrenergic characteristics 3'4. Such an explanation appears quite possible, as it has been found that sudomotor neurons develop their cholinergic properties postnatally, and retain some characteristics of their former noradrenergic phenotype 16"w. In the experiments where fluorogold was injected into the paw pads, capillary endothelial cells were labelled as well as neurons. Presumably the endothelial ceils took up tracer which had leaked into the blood stream, because whichever paw had been injected, endothelial cells were labelled in all the ganglia studied ipsilateral, contralateral, stellate and lumbar. By contrast, neuronal labelling was seen only in a minority of cells, and only in the ganglion whose neurons supplied the injected paw. We are therefore confident that the labelling of neurons with fluorogold resulted from retrograde axonal transport. Two known populations of sympathetic neurons project to the paw pad: sudomotor neurons and the noradrenergic vasoconstrictor neurons which supply skin vessels. Probably only a minority of those populations were detected by retrograde labelling, partly because the process is not 100% efficient, and also because the intensity of fluorogold fluorescence was reduced by the immunohistochemical procedure, so some faintly labelled cells could have been missed. But because the same considerations apply to both classes of neurons. their relative proportions were probably preserved among the fluorogold labelled cells. Approximately, one third of these were surrounded by CRF-LI terminal baskets. The true proportion of sudomotor neurons may be slightly higher than this figure, because the presence of CRF-LI terminal baskets identifies only 84% of presumptive sudomotor n e u r o n s ( s e e above l. -

Chemical coding for function among preganglionic neurons Other studies on guinea pigs and rats have begun to provide data on the likely functions of preganglionic neurons with a particular staining pattern. Recently, Gibbins 1° used earlier data on the histochemistry and functional targets of postganglionic neurons to classify the preganglionic neurons which innervate them. He found in guinea pig lumbar ganglia that preganglionic fibres containing CGRP-LI selectively terminated on a subset of putative vasoconstrictor neurons, that those containing SP-LI terminated on putative vasodilator

259 neurons, and that those containing VIP-LI terminated on putative pilomotor neurons ~°. In the rat, Forehand 5'6 found that about 27% of stellate ganglion cells which had been retrogradely labelled from the paw, but virtually none of those labelled retrogradely from the interscapular brown fat, were surrounded by terminals containing enkephalin (ENK)-LI. ENK-LI may thus be a label for the preganglionic sudomotor neurons in rats. It may also be a label for preganglionic sudomotor neurons in cats, but it is unlikely to be a specific label: there, terminals with ENK-LI are found around virtually all cholinergic ganglion cells (i.e. vasodilator as well as sudomotor neurons) 2~. Whether the same applies or not to rats is unclear. Because the chemical code for function may differ between species 7, this discussion will now focus on data from the cat.

Chemical coding in feline paravertebral ganglia Immunoreactivity for peptides and peptide combinations has been shown to label several subsets of preganglionic neurons in the cat 12-14, but what functions each subset might subserve were unknown. In the paravertebral ganglia, Kummer and Heym ~5 identified 5 histochemical types of fibre terminals, and showed that each selectively innervated particular histochemical types of ganglion cells. Of particular interest here are the ganglion cells that contain CGRP-LI, of which they found many in the superior cervical ganglion ~5. These cells were contacted either by terminals containing somatostatin (SOM)-LI or by terminals containing dopamine/3 hydroxylase (DBH)-LI in combination with either ENK-LI or neurotensin (NT)-LI. Presumably the neurons in that ganglion supply targets in the head rather than the paws, so these pathways are not sudomotor in function. However, the same pre-postganglionic conjunctions were stated to occur in the stellate ganglion 15, where neurons with CGRP-LI have been identified as sudomotor 2°. Of the preganglionic markers in question here, SOM-LI was found not to be specific to any one pathway; but DBH-LI, with either ENK- or NT-LI, was stated to stain terminals only around cells showing CGRP-L115. It is not clear exactly what DBH-LI in that study represents, however, because cat preganglionic neurons are not noradrenergic, nor have any terminal structures in the stellate ganglion been found by ourselves (see above) or others 2° to express immunoreactivity for TH, the immediate precursor enzyme for catecholamine synthesis. Alone, neither ENK-LI nor NT-LI appear to be specific markers of preganglionic sudomotor neurons in the cat: terminals with ENK-LI also innervate va-

sodilator neurons 2~, while terminals with NT-LI also innervate noradrenergic cells in thoracic and lumbar ganglia 15,zl. The inference that those immunoreactivity patterns are not specific to the sudomotor pathway is also born out by their segmental distribution: both ENK-LI and NT-LI have been found in preganglionic cell bodies at all thoracolumbar spinal segments 13'14. By contrast, preganglionic neurons with CRF-LI are concentrated only in segments T 2 - T 7 and L 2- L ~2 3, reinforcing the notion that they may be of a single functional type.

Functional significance of CRF-LI What functional role CRF or a CRF-like substance may play in preganglionic sudomotor neurons remains to be determined. It might act as a peptide co-transmitter with acetylcholine. Indeed, a peptide-like substance appears to play a co-transmitter role in the cardioaccelerator pathway through the cat stellate ganglion 2, but corresponding data for the sudomotor pathway are lacking. Interestingly, prolonged 40 Hz stimulation of preganglionic fibres exhausted the presumptively peptidergic cardioaccelerator response and depleted the ENK-LI and NT-LI, but not the CRF-, SP-, SOM- or VIP-LI found in preganglionic terminals of the cat stellate ganglion ~. CONCLUSION CRF-LI is probably a specific, though not necessarily universal, marker of preganglionic sudomotor neurons in the cat. Acknowledgements. We are grateful to John Furness, Heather Young and Daphne Hards for advice and help during this study, which was supported by the National Health and Medical Research Council of Australia. The CRF antibody was kindly given by Dr. W. Vale of the Salk Institute. A.D.S. is a Howard Florey Institute Postgraduate Scholar.

REFERENCES 1 Bachoo, M., Ciriello, J. and Polosa, C., Effects of preganglionic stimulation on neuropeptide-like immunoreactivity in the stellate ganglion of the cat, Brain Res., 400 (1987) 377-382. 2 Bachoo, M., Isacoff, E. and Polosa, C., Slow cardioacceleration mediated by noncholinergic transmission in the stellate ganglion of the cat, Can. J. Physiol. Pharmacol., 66 (1988) 1066-1074. 3 Burnstock, G., Do some nerve cells release more than one transmitter?, Neuroscience, 1 (1976) 239-248. 4 Burnstock, G., Do some sympathetic neurones synthesize and release both noradrenaline and acetylcholine? Prog. Neurobiol., II (1987) 205-222. 5 Forehand, C.J., Peptidergic specificty in the innervation of rat sympathetic ganglion cells, Anat. Rec., 223 (1989) 394. 6 Forehand, C.J., Development of specific, leu-enkephalinergic input to rat stellate ganglion cells that project to sweat glands, J. Auton. Neru. Syst., 33 (1991) 217-218. 7 Furness, J.B., Morris, J.L., Gibbins, I.L. and Costa, M., Chemical

260

8 9

10

11

12

13 14

15

16

17

coding of neurones and plurichemical transmission, Annu. Rez'. Pharmacol. Toxicol., 29 (1989) 289-306. Gabella, G., Structure of the Autonornk: Ner~,ous System, Chapman, London, 1976. Gibbins, I.L., Furness, J.B., Costa, M., MacIntyre, I., Hillyard, C.J. and Girgis, S., Co-localization of calcitonin gene-related peptide-like immunoreactivity with substance P in cutaneous, vascular and visceral sensory neurones of guinea pigs, Neurosci. Lett., 57 (1985) 125-130. Gibbins, I.L., Vasoconctrictor, vasodilator and pilomotor pathways in sympathetic ganglia of guinea-pigs, Neuroscience, 47 (1992) 657-672. J~inig, W., Organization of the lumbar sympathetic outflow to skeletal muscle and skin of the cat hindlimb and tail, Ret,. Physiol. Biochem. Pharmaeol., 102 (1985) 119-213. Krukoff, T.L., Segmental distribution of corticotropin-releasing factor-like and vasoactive intestinal peptide-like immunoreactivities in presumptive sympathetic preganglionic neurons of the cat, Brain Res., 382 (1986) 153-157. Krukoff, T.L., Coexistence of neuropeptides in sympathetic preganglionic neurons of the cat, Peptides, 8 (1987) 109-112. Krukoff, T.L., Ciriello, J. and Calaresu, F.R., Segmental distribution of peptide-like immunoreactivity in cell bodies of the thoracolumbar sympathetic nuclei of the cat, J. Comp. Neurol., 240 (1985) 90-102. Kummer, W. and Heym, C., Neuropeptide distribution in the cervico-thoracic paravertebral ganglia of the cat with particular reference to calcitonin gene-related peptide immunoreactivity, Cell Tissue Res., 252 (1988) 463-471. Landis, S.C., Development of cholinergic sympathetic neurons: evidence for transmitter plasticity in vivo, Fed. Proc., 2 (1983) 1633 1638. Landis, S.C., Neurotransmitter plasticity in sympathetic neurons and its regulation by enviromnental factors in vitro and in vivo. In A. Bj6rklund, T. H6kfelt and C. Owman (Eds.), Handbook of

18

19

20

21

22

23

24

25

26

Chemical Neuroanatomy, Vol. 6, Elsevier. Amsterdam. 1988. pp. 65-115. Langley, J.N., On the course and connections of the secretory fibres supplying the sweat glands on the feel t~[ the caT..! Physiol., 12 (189l) 347-374. Langley, J.N., Further observations on the ~,ccrctory and vasomotor fibres of the foot of the cat, with notes on other sympathetic nerve fibres, J. Physiol., 17 (1895) 296-314. Lindh, B., Lundberg, J.M. and H6kfelt, T., NPY-, VIP/PhI-, CGRP- and substance P-immunoreactive neuronal subpopulations in cat autonomic and sensory ganglia and their projections, Cell Tissue Res., 256 (1989) 259-273. Lindh, B., Pelto-Huikko, M., Risling, M. and H6kfelt, T., Cholinergic sympathetic neurones, J. Auton~ Nerl,. ,~'st., 33 (1991) 106107. Lundberg, J. and Tatemoto, K., Pancreatic polypeptide family (APP, BPP, NPY and PYY) in relation to sympathetic vasoconstriction resistant to a-adrenoreceptor blockade, Acta Physiol. Scand., 116 (1982) 393-402. Potter, E., Prolonged non-adrenergic inhibition of cardiac vagal action following sympathetic stimulation: neuromodulation by neuropeptide Y? Neurosci. Lett., 54 (1985) 117-121. Schultzberg, M. and Lindh, B., Transmitters and peptides in autonomic ganglia. In A. Bj6rklund, T. H6kfelt and C. Owman (Eds.), Handbook of Chemical Neuroanatomy, VoL 6, Elsevier, Amsterdam, 1988, pp. 397-426. Shafton, A.D., Oldfield, B.J. and McAIlen, R:M., CRF: a marker for sympathetic preganglionic neurons which innervate nonnoradrenergic (presumably cholinergic) ganglion cells in cats, Proc. Aust. Neurosci. Sot., 3 (1992) 117. Watabe, T., Levidiotis, M.L., Oldfield, B. and Wintour, E.M., Ontogeny of corticotropbin-releasing factor (CRF) in the ovine fetal hypothalamus: use of multiple CRF antibodies, .L Endocrinol., 129 (199l) 335-341.